Characterizing the Impact of Doppler Effects on Body-Centric Lora Links with SDR
Total Page:16
File Type:pdf, Size:1020Kb
sensors Article Characterizing the Impact of Doppler Effects on Body-Centric LoRa Links with SDR Thomas Ameloot 1,* , Marc Moeneclaey 2 , Patrick Van Torre 1 and Hendrik Rogier 1 1 IDLab, Department of Information Technology (INTEC), Ghent University-imec, Technologiepark-Zwijnaarde 126, B-9052 Ghent, Belgium; [email protected] (P.V.T.); [email protected] (H.R.) 2 Department of Telecommunications and Information Processing (TELIN), Ghent University, Sint-Pietersnieuwstraat 41, B-9000 Ghent, Belgium; [email protected] * Correspondence: [email protected]; Tel.: +32-9-331-4881 Abstract: Long-range, low-power wireless technologies such as LoRa have been shown to exhibit excellent performance when applied in body-centric wireless applications. However, the robustness of LoRa technology to Doppler spread has recently been called into question by a number of researchers. This paper evaluates the impact of static and dynamic Doppler shifts on a simulated LoRa symbol detector and two types of simulated LoRa receivers. The results are interpreted specifically for body-centric applications and confirm that, in most application environments, pure Doppler effects are unlikely to severely disrupt wireless communication, confirming previous research, which stated that the link deteriorations observed in a number of practical LoRa measurement campaigns would mainly be caused by multipath fading effects. Yet, dynamic Doppler shifts, which occur as a result of the relative acceleration between communicating nodes, are also shown to contribute to link degradation. This is especially so for higher LoRa spreading factors and larger packet sizes. Keywords: Internet of Things; LoRa; body-centric communication; software defined radio; doppler Citation: Ameloot, T.; Moeneclaey, M.; Van Torre, P.; Rogier, H. Characterizing the Impact of Doppler Effects on Body-Centric LoRa Links 1. Introduction with SDR. Sensors 2021, 21, 4049. In recent years, sub-GHz low-power wide-area network (LPWAN) technologies such https://doi.org/10.3390/s21124049 as SigFox [1], NB-IoT [2], and LoRa [3] have played key roles in the development of the rapidly evolving Internet of Things (IoT). Following the widespread adoption of these Academic Editor: Lorenzo Vangelista technologies in a wide range of application environments, several research efforts have been devoted to assessing the viability of using LoRa modulation for body-centric wireless Received: 05 May 2021 communication [4–6]. LoRa employs wide-band frequency-modulated pulses called chirps Accepted: 07 June 2021 to achieve spreading gain, which results in the successful reception of packets at extremely Published: 12 June 2021 low signal-to-noise ratio (SNR) levels [7]. Compared to its competitors, LoRa is especially suitable for body-centric applications as its data rate can be adapted, which has been Publisher’s Note: MDPI stays neutral shown to benefit the coverage of body-centric LoRa networks [8]. LoRa modulation is with regard to jurisdictional claims in most often deployed in the 868 MHz industrial, scientific and medical (ISM) radio band. published maps and institutional affil- iations. For body-centric wireless communication, this band is very interesting as it both enables applications to benefit from the excellent propagation characteristics observed at sub-GHz frequencies, while still allowing compact wearable antennas to be designed, given the wavelength of 35 cm. To properly assess the viability of using LoRa modulation in body-centric networks, it Copyright: © 2021 by the authors. is important to thoroughly evaluate its physical layer performance. Unfortunately, as LoRa Licensee MDPI, Basel, Switzerland. is a proprietary technology, research into its wireless performance has been challenging. This article is an open access article Until now, it has mostly relied on either theoretical reviews [9–12] or channel measurements distributed under the terms and conditions of the Creative Commons gathered by commercial transceivers [4–7]. Recently, efforts have been carried out to Attribution (CC BY) license (https:// implement LoRa modulation in signal processing code, either for simulation purposes or creativecommons.org/licenses/by/ for implementation on software-defined radio (SDR) platforms [13–18]. Having access 4.0/). to individual I and Q samples enables much more accurate channel estimation than is Sensors 2021, 21, 4049. https://doi.org/10.3390/s21124049 https://www.mdpi.com/journal/sensors Sensors 2021, 21, 4049 2 of 15 possible with commercial transceivers. Furthermore, using these implementations, different propagation mechanisms can be simulated, and their influence on LoRa modulation can be analyzed. One mechanism that can impact body-centric wireless links is the Doppler effect. As people move around, their relative velocities change continuously. Several sources declare that LoRa shows good immunity against the Doppler effect [19,20]. Some practical studies confirm this, however, others do not. In [21], LoRa link degradations are demonstrated for relative velocities around 40 km/h. In [22], the Doppler effect is blamed for severe link degradations. In [23], the authors presented two body-to-base-station measurement campaigns at different velocities (6.2 km/h and 31.1 km/h), which show no significant dif- ference. These contradictions were also examined in [24], which presents lab measurements and outdoor experiments investigating the Doppler effect in LoRa satellite communication. In [24], it is demonstrated that LoRa is indeed reasonably Doppler-resistant, and it is stated that the link degradations observed in [21] are expected to be the result of Doppler spread. However, most of these conclusions are based on fully experimental examinations. This paper assesses the impact of Doppler effects on a recently developed SDR imple- mentation of LoRa [25]. General conclusions are drawn based on computer simulations of static and dynamic Doppler effects, assuming a worst-case angle of incidence. For both cases, a comparison is made between two packet synchronization strategies. Results are also interpreted specifically for body-centric LoRa networks. Finally, guidelines are provided on how static and dynamic Doppler effects can be mitigated in LoRa networks, e.g., by modifying the packet structure. The paper is structured as follows. In Section2, LoRa modulation is described in general. Additionally, key points on the SDR implementa- tion applied in this work are presented. Section3 elaborates on relevant considerations published in other research and describes the software-based simulation of Doppler effects on LoRa modulation. A discussion of the results in perspective of previous research is presented in Section4. General conclusions are drawn in Section5. 2. SDR-Based LoRa Modulation 2.1. LoRa Modulation Basics LoRa is based on chirp spread spectrum (CSS) modulation [26], which uses wide-band frequency-modulated pulses to encode information. The most important modulation param- eters are the spreading factor SF 2 f7, 8, 9, 10, 11, 12g, bandwidth BW 2 f125, 250, 500g kHz, and code rate CR 2 f4/5, 4/6, 4/7, 4/8g. In most LoRa research, the bandwidth is actually used to describe the frequency swing B of the signals, as very little energy is present for frequencies outside of the range described by B. The spreading factor determines the slope of the chirp w.r.t. the frequency swing. For a given B, a higher spreading factor spreads any given LoRa symbol over a longer symbol interval. A detailed description of LoRa packet air times and data rates for different SF values is provided in [8]. Increasing the spreading factor or the frequency swing also impacts the sensitivity of LoRa receivers, as presented in [27]. On the packet level, a default preamble length of 12.25 symbols is considered. This preamble consists of eight up-chirps that constitute the pilot sequence, two so-called sync word symbols, which enable the user to distinguish packets from different LoRa networks, and 2.25 down-chirps, which make up the start-of-frame delimiter (SFD), used for packet synchronization. The optional header and/or cyclic redundancy check that can be added to LoRa packets are considered to be part of the data payload. The interleaving and decoding steps that are applied to data when encapsulated in a LoRa packet have been documented in [13] and are considered to be part of the data link layer. Consequently, these are not discussed in this paper. Additional details and considerations on high-level aspects of LoRa and LoRaWAN can be found in [27,28]. 2.2. Software Implementation As mentioned earlier, this paper employs a software implementation of LoRa to investigate the impact of Doppler shifts on the physical layer of LoRa modulation. In this Sensors 2021, 21, 4049 3 of 15 implementation, which is based on LoRa waveform expressions presented in [12,25], a discrete LoRa up-chirp with symbol energy Es and symbol duration Ts is described by the waveform r E a T2n2 s [n] = s exp j 2p s , (1) up N 2 N2 where n 2 f0, 1, 2, ..., N−1g signifies the sample index and N equals the amount of samples used to describe the up-chirp. N is related to the spreading factor through N = K · M, SF with M = 2 and K 2 Z+ the oversampling factor. The slope a (in Hz/s) of the up-chirp is related to the symbol duration and the frequency swing B through a = B/Ts. A LoRa symbol a 2 f0, 1, 2, ..., M−1g is encoded by changing the starting frequency of the chirp to aB/M and resetting the instantaneous frequency of the chirp to zero when it reaches B. According to the slope of the chirp, this occurs at the time instant when n = (1 − a/M)N. After the reset, the chirp frequency again increases linearly according to a. Thus, the encoded LoRa symbol s[n, a] is described by r E a T2n2 T n s[n, a] = s exp j 2p s + F [n, a] s , (2) N 2 N2 a N ( aB n < (1− a )N [ ] = M M where Fa n, a aB a .